SUMMARY
Genetic selection of farm animals for productivity, and intensification of farming practices have yielded substantial improvements in efficiency; however, the capacity of animals to cope with environmental challenges has diminished. Understanding how the animal and environment interact is central to efforts to improve the health, fitness, and welfare of animals through breeding and management strategies. The review examines aspects of the environment that are sensed by the animal. The predictive brain model of sensory perception and motor action (the Bayesian brain model) and its recent extension to account for anticipatory, predictive control of physiological activities is described. Afferent sensory input from the immune system, and induction of predictive immune functions by the efferent nervous system are also in accord with the Bayesian brain model. In this model, expected sensory input (from external, physiological and immunological environments) is reconciled with actual sensory input through behavioural, physiological and immune actions, and through updating future expectations. Sustained discrepancy between expected and actual sensory inputs occurs when environmental encounters cannot be predicted and controlled resulting in stress and negative affective states. Through these processes, from the animal’s perspective, aspects of the environment acquire a negative or positive character: that is the environment becomes valenced. In a homeostatic manner, affective experience guides the animal towards synchronisation and a greater degree of mutualism with its environment. A better understanding of the dynamic among environmental valence, animal affect and mutualism may provide a better understanding of genetic and phenotypic links between temperament, immune function, metabolic performance, affective state, and resilience in farm animals, and provide further opportunities to improve their welfare.